Main
Lisa Oshita
I am a blah blah blah seeking blah blah blah
Education
B.S. Statistics
California Polytechnic State University
San Luis Obispo, CA
Dec 2018 - Sept 2014
Selected Positions
Data Science Intern
Brandless, Inc.
San Francisco, CA
Sept 2018 - June 2018
- Implemented email marketing strategy to target unconverted customers with specific promotions, based on random forest predictions of average order value
- Developed classifier to predict retention based on early shopping behavior
- Worked on box recommendation algorithm to minimize empty space and two-box shipments to enable cost savings of > $100,000
- Performed text mining, sentiment analysis on customer feedback to derive insights on overall trends in opinion
Undergraduate Researcher
iFixit, Cal Poly Department of Statistics
San Luis Obispo, CA
Sept 2017 - June 2017
- Modeled response times of questions on iFixit’s Q&A forum, using Cox regression, to predict question survival probabilities, identify signficantly associated factors
- Implemented Shiny app to allow users to compute survival probabilities for any set of questions
Data Scientist, Analytics
Brandless, Inc.
San Francisco, CA
Feb 2020 - Jan 2019
- Advised A/B tests, investigated alternative experimentation methods, e.g. Non-Inferiority Tests, for optimized, faster testing
- Constructed, improved Redshift tables, Looker dashboards, metrics, backend infrastructure to support both data and external team functions
- Performed deep-dive analyses into user-site behavior to guide product decisions
- Worked with Marketing to identify profitable, loyal customers to target with email campaigns, product recommendations and promotions based on segmentation and order history
- Developed models to predict interest in CBD, perceived CBD efficacy and retention based on preferences and demographic information obtained from survey responses, to inform CBD Product Finder Quiz design
Data Science Intern
Air Pollution Control District
San Luis Obispo, CA
Dec 2018 - Mar 2018
- Implemented R, Python scripts to automate AirAware text alerts sent to local residents
- Optimized alert system with cumulative, seasonally-varying threshold algorithm to proactively trigger texts
- Performed deep-dive analyses into user-site behavior to guide product decisions
- Created R package to load, format, analyze Federal Air Quality Systems data